20.500.12556/DKUM-90165 Razpoznava objektov ter napovedovanje njihovega obnašanja v aplikacijah za avtonomno vožnjo magistrsko delo Object recognition and behavior prediction in autonomous driving applications V magistrskem delu smo se osredotočili na raziskovanje in implementacijo sistema za avtonomno vožnjo, ki služi kot osnova za sprejemanje odločitev. Informacije iz okolice vozila pridobivamo zgolj s pomočjo kamere. Za implementacijo smo uporabili dve vrsti kamer, in sicer monokularno ter stereo. Nad podatki monokularne kamere smo implementirali algoritme za prepoznavanje objektov v okolici vozila, sledenje tem objektom skozi zaporedje slik, zaznavanje voznega pasu, določanje pozicije vozila na njem in ocenjevanje časa do roba vozišča. Podatke stereo kamere smo uporabili za določanje razdalje do prepoznanih objektov ter konstrukcijo 3D strukture okolice premikajočega se vozila. In the master's thesis, we focused on researching and implementing a system for autonomous driving, serving as a foundation for decision-making. Information from the vehicle's surroundings is obtained solely through the use of a camera. For the implementation, we utilized two types of cameras, namely monocular and stereo. Using the data from monocular camera, we implemented algorithms for object recognition in the vehicle's surroundings, tracking these objects through a sequence of images, detecting the driving lane, determining the vehicle's position on it, and estimating the time to the lane's edge. Data from the stereo camera were employed for determining the distance to recognized objects and constructing a 3D structure of the moving vehicle's surroundings. avtonomna vožnja stereo vid računalniški vid konvolucijske nevronske mreže autonomous driving stereo vision computer vision convolutional neural networks true true false [N. Kolar] Slovenski jezik Angleški jezik Magistrsko delo/naloga 2024-08-23 13:59:12 2024-10-14 09:48:30 2024-12-18 03:09:05 0000-00-00 00:00:00 2024 0 Maribor 0 Maribor 1 spletni vir (1 datoteka PDF (VIII, 89 f.)) 0000-00-00 NiDoloceno NiDoloceno NiDoloceno 0000-00-00 0000-00-00 0000-00-00 004.352.242:629.4.072(043.2) 219372803 MAG_Kolar_Niko_2024.pdf MAG_Kolar_Niko_2024.pdf 1 FC668A8C095C525586B45391844728FE dc0b9eb63f43e60a264de21bb3198247d9ebb20acf196b004522c1dc90550360 b795633a-62f0-11ef-a652-00155d000105 https://dk.um.si/Dokument.php?lang=slv&id=183261 Fakulteta za elektrotehniko, računalništvo in informatiko 0 0 0